CN110321214A

CN110321214A - A kind of data query method, device and equipment

Info

Publication number: CN110321214A
Application number: CN201810268968.0A
Authority: CN
Inventors: 周祥; 李冰; 赵永春; 温绍锦
Original assignee: Alibaba Group Holding Ltd
Current assignee: Alibaba Group Holding Ltd
Priority date: 2018-03-29
Filing date: 2018-03-29
Publication date: 2019-10-11
Also published as: JP2021519460A; EP3779688A4; US11556541B2; US20210011916A1; WO2019184739A1; JP7511477B2; EP3779688A1

Abstract

The application provides a kind of data query method, device and equipment, this method comprises: obtaining resource overhead according to the characteristic information of the inquiry request received；According to the calculate node in the resource overhead and calculate node resource dynamic adjustment resource pond；Data corresponding with the inquiry request are inquired by the calculate node.Pass through the technical solution of the application, it can be with the calculate node in dynamic adjustment resource pond, so that the calculate node in resource pool is capable of handling all inquiry requests received, more effectively improve the treatment effeciency and resource utilization of calculate node, calculate node more effectively parallel processing multiple queries are requested, improve cpu resource, memory source, network bandwidth resources utilization rate, to reach a better effect from overall calculation resource and user query angles of loading, user's use feeling is improved.

Description

A kind of data query method, device and equipment

Technical field

This application involves Internet technical field more particularly to a kind of data query method, device and equipments.

Background technique

Open Analysis Service (Open Analytics) is used to provide looking into for serverless backup (Serverless) for user Analysis Service is ask, the analysis and inquiry of any dimension can be carried out to mass data, supports that (Millisecond is rung for high concurrent, low delay Answer), on-line analysis, mass data inquiry etc. functions.It may include data source and calculating section in open Analysis Service system Point, data source for storing mass data, inquire from data source and ask with the inquiry after receiving inquiry request by calculate node Seek corresponding data.

But under certain application scenarios (inquiry scene, the inquiry scene of representation data of such as map datum), calculate Node may receive multiple queries request (i.e. number of concurrent is very high) in a short time, that is, need to handle multiple look into a short time Request is ask, CPU (Central Processing Unit, central processing unit) resource, memory source, network bandwidth etc. is caused to go out It is now abnormal, so as to cause query timeout or inquiry failure.

Summary of the invention

The application provides a kind of data query method, which comprises

Resource overhead is obtained according to the characteristic information of the inquiry request received；

According to the calculate node in the resource overhead and calculate node resource dynamic adjustment resource pond；

Data corresponding with the inquiry request are inquired by the calculate node.

The application provides a kind of data query method, which comprises

According to the characteristic information of the inquiry request received, the inquiry request received is divided at least one distribution Group；Wherein, different distribution groups corresponds to different child resource ponds；

The resource overhead of the distribution group is obtained according to the characteristic information of the inquiry request in distribution group；

According to the calculate node resource of the resource overhead of the distribution group and the corresponding child resource pond of the distribution group, dynamic Adjust the calculate node in the child resource pond；

Data corresponding with the inquiry request in the distribution group are inquired by the calculate node in child resource pond.

The application provides a kind of data query device, and described device includes:

Module is obtained, for obtaining resource overhead according to the characteristic information of the inquiry request received；

Processing module, according to the calculate node in resource overhead and calculate node resource dynamic adjustment resource pond；

Enquiry module, for inquiring data corresponding with the inquiry request by the calculate node.

The inquiry request received is divided by division module for the characteristic information according to the inquiry request received At least one distribution group；Wherein, different distribution groups correspond to different child resource ponds；

Module is obtained, the resource for obtaining the distribution group according to the characteristic information of the inquiry request in distribution group is opened Pin；

Processing module, for according to the resource overhead of the distribution group and the calculating in the corresponding child resource pond of the distribution group Node resource, the calculate node in child resource pond described in dynamic regulation；

Enquiry module, for passing through the calculate node inquiry in the child resource pond and the inquiry request in the distribution group Corresponding data.

The application provides a kind of data query equipment, comprising: processor, for the feature according to the inquiry request received Information acquisition resource overhead；According to the calculate node in resource overhead and calculate node resource dynamic adjustment resource pond；Pass through institute It states calculate node and inquires data corresponding with the inquiry request.

The application provides a kind of data query equipment, comprising: processor, for the feature according to the inquiry request received The inquiry request received is divided at least one distribution group by information；Wherein, different distribution groups corresponds to different child resources Pond；The resource overhead of the distribution group is obtained according to the characteristic information of the inquiry request in distribution group；According to the distribution group The calculate node resource of resource overhead and the corresponding child resource pond of the distribution group, the calculating in child resource pond described in dynamic regulation Node；Data corresponding with the inquiry request in the distribution group are inquired by the calculate node in child resource pond.

Based on the above-mentioned technical proposal, it in the embodiment of the present application, can be obtained according to the characteristic information of the inquiry request received Resource overhead is obtained, and according to the calculate node in resource overhead and calculate node resource dynamic adjustment resource pond, so that resource pool In calculate node be capable of handling all inquiry requests received, more effectively improve calculate node treatment effeciency and resource Utilization rate, can enable calculate node more effectively parallel processing multiple queries request, improve cpu resource, memory source, The utilization rate of network bandwidth resources, thus reach a better effect from overall calculation resource and user query angles of loading, Improve user's use feeling.Moreover, allowing each calculate node to be user by the calculate node in dynamic adjustment resource pond The query analysis service of serverless backup (Serverless) is provided, so that user is not necessarily to aware services device or Service Instance, The service of cloud service offer itself need to be only provided, cloud service is based on, user only needs to input SQL query request, so that it may by counting Operator node carries out data query and analysis in data source, can be with Seamless integration- business analysis tool and application program.It can be right Resource carries out intellectual analysis and adjust automatically, more effectively improves the utilization of resources of cloud database and cloud data analysis service cluster Rate and cost performance.

Detailed description of the invention

It, below will be to the application in order to clearly illustrate the embodiment of the present application or technical solution in the prior art Embodiment or attached drawing needed to be used in the description of the prior art be briefly described, it should be apparent that, it is described below Attached drawing is only some embodiments as described in this application, for those of ordinary skill in the art, can also be according to this Shen Please these attached drawings of embodiment obtain other attached drawings.

Fig. 1 is the system structure diagram in a kind of embodiment of the application；

Fig. 2 is the flow chart of the data query method in a kind of embodiment of the application；

Fig. 3 is the system structure diagram in the application another embodiment；

Fig. 4 is the flow chart of the data query method in the application another embodiment；

Fig. 5 is the structure chart of the data query device in a kind of embodiment of the application；

Fig. 6 is the structure chart of the data query device in the application another embodiment.

Specific embodiment

In the term that the embodiment of the present application uses merely for the sake of for the purpose of describing particular embodiments, rather than limit this Shen Please.The "an" of singular used in the application and claims, " described " and "the" are also intended to including most shapes Formula, unless context clearly shows that other meanings.It is also understood that term "and/or" used herein refers to comprising one A or multiple associated any or all of project listed may combine.

It will be appreciated that though various letters may be described using term first, second, third, etc. in the embodiment of the present application Breath, but these information should not necessarily be limited by these terms.These terms are only used to for same type of information being distinguished from each other out.For example, In the case where not departing from the application range, the first information can also be referred to as the second information, and similarly, the second information can also be with The referred to as first information.Depending on context, in addition, used word " if " can be construed to " ... when " or " when ... " or " in response to determination ".

The embodiment of the present application proposes that a kind of data query method, this method can be applied to include client, load balancing Equipment, front end node (front node, be referred to as front-end server), (compute node can also claim calculate node For calculation server) and data source system, such as system for realizing open Analysis Service.It is, of course, also possible to include other Server, such as resource allocation server, it is without limitation.

It is shown in Figure 1, it is the application scenarios schematic diagram of the embodiment of the present application, in the resource pool of front end node, including One or more front end nodes, Fig. 1 is by taking 3 front end nodes as an example.In the resource pool of calculate node, including one or more meters Operator node, Fig. 1 is by taking 5 calculate nodes as an example.In practical application, dilatation can be carried out to front end node and (increase front end node Quantity) or capacity reducing (quantity for reducing front end node), dilatation can also be carried out to calculate node and (increase the number of calculate node Amount) perhaps capacity reducing the quantity of calculate node (reduce) the present embodiment be exactly the side for being directed to calculate node and carrying out dilatation or capacity reducing Case.

Wherein, client such as can be terminal device (such as PC (Personal for inquiring data from data source Computer, personal computer), laptop, mobile terminal etc.) APP (Application, using) that includes, can also be with It is the browser that terminal device includes, the type of this client is not particularly limited.

Wherein, load-balancing device is used to carry out load balancing to inquiry request, for example, receiving a large amount of inquiry requests It afterwards, can be by these inquiry request load balancing to each front end node, with no restrictions to this process.

Wherein, data source is for storing various types of data, and the data that stored in data source can be supplied to visitor Family end.Type for the data stored in data source in the embodiment of the present application with no restrictions such as can be user data, quotient Product data, map datum, video data, image data, audio data etc..

Wherein, multiple front end nodes in resource pool are for providing identical function.Specifically, front end node is for receiving The inquiry request that client is sent, and SQL (Structured Query Language, structuring are carried out to the inquiry request Query language) parsing, inquiry request is generated using SQL parsing result, and the inquiry request is sent to calculate node, the inquiry Request is for requesting data corresponding with the inquiry request.Then, front end node is also used to receive the data of calculate node return, And send the data to client.

Wherein, multiple calculate nodes in resource pool are for providing identical function.Specifically, calculate node is for receiving The inquiry request that front end node is sent, and data corresponding with the inquiry request are read from data source using the inquiry request, With no restrictions to this reading process, and front end node is sent the data to.

In one example, if calculate node receives a large amount of inquiry requests (i.e. number of concurrent is very high) in a short time, Calculate node needs to handle a large amount of inquiry requests in a short time, causes cpu resource, memory source, network bandwidth etc. to occur different Often, so as to cause query timeout or inquiry failure.Unlike aforesaid way, in the embodiment of the present application, can dynamically it adjust Calculate node in section resource pool can increase the calculate node quantity in resource pool, subtract that is, when there are a large amount of inquiry requests The inquiry request quantity of few each calculate node, so that some calculate node is avoided to handle a large amount of inquiry requests in a short time, The treatment effeciency and resource utilization for more effectively improving calculate node mitigate cpu resource, memory source, network bandwidth and account for With, process performance can be improved, and avoid client query time-out or failure, improve user's use feeling.

It is shown in Figure 2 under above-mentioned application scenarios, for the process of the data query method proposed in the embodiment of the present application Schematic diagram, this method can be applied to data query equipment, the data query equipment can in Fig. 1 load-balancing device, Or front end node or resource allocation server, it is without limitation, in the present embodiment for being applied to front end node, This method may comprise steps of:

Step 201, resource overhead is obtained according to the characteristic information of the inquiry request received.For example, can be according to default The characteristic information of the inquiry request received in time window obtains resource overhead.

Step 202, according to the calculate node in the resource overhead and calculate node resource dynamic adjustment resource pond.

Step 203, data corresponding with the inquiry request are inquired by the calculate node in resource pool.

In one example, above-mentioned execution sequence is intended merely to facilitate description to provide example, in practical applications, Sequence is executed between can also changing the step, with no restrictions to this execution sequence.Moreover, in other embodiments, and it is different The fixed sequence for showing and describing according to this specification is come the step of executing correlation method, step included by method can be than this It is more or less described in specification.In addition, single step described in this specification, it in other embodiments may quilt Multiple steps are decomposed into be described；Multiple steps described in this specification may also be merged into other embodiments Single step is described.

Wherein, when client needs the data in request data source, inquiry request can be sent, load-balancing device exists After receiving the inquiry request, which can be sent to front end node, front end node is receiving the inquiry request Afterwards, which can be stored in query request (Query Queue).

Wherein, preset time window can be set in front end node, and the time of the preset time window can rule of thumb configure, such as 3 seconds etc..Based on this, all inquiry requests being stored in query request in preset time window can be determined as by front end node The inquiry request received in preset time window, such as 100 inquiry requests.

Before executing step 201, for all inquiry requests received in preset time window, can also first it obtain every The characteristic information of a inquiry request, this feature information can include but is not limited to following one or any combination: concurrency is looked into Ask complexity, inquiry data volume, query time, resources occupation rate.

One, concurrency (Concurrency), i.e., received inquiry request quantity in preset time window, such as 100.

Two, complexity (Query_Complexity) is inquired, that is, executes the complexity of inquiry request, can indicates to inquire Situations such as time, cpu resource occupy, memory source occupies, network bandwidth occupies.Wherein, inquiry complexity is usually a number Value can be looked by query time, cpu resource occupancy, memory source occupy, network bandwidth is occupied and is normalized Ask the numerical value of complexity.For example, if needing to occupy a large amount of cpu resources, memory source, network bandwidth when executing inquiry request 1, Query time is longer, then the inquiry complexity of inquiry request 1 is higher.If execute inquiry request 2, need to occupy a small amount of CPU money Source, memory source, network bandwidth, query time is shorter, then the inquiry complexity of inquiry request 2 is lower.

For the inquiry request with same queries keyword, inquiry complexity is same or similar, therefore, can obtain Take the corresponding relationship of key word of the inquiry and complexity value, and in the first mapping table record queries keyword and complexity value pair It should be related to.For example, it is assumed that inquiry request 1 and inquiry request 2 are the inquiry requests for key word of the inquiry A, then inquiry request 1 It is identical with the inquiry complexity of inquiry request 2.Assuming that in the first mapping table record queries keyword A and complexity value A pair It should be related to, then for inquiry request 1 and inquiry request 2, the inquiry complexity of inquiry request 1 and inquiry request 2 is multiple Miscellaneous angle value A.

Wherein, the corresponding relationship for obtaining key word of the inquiry and complexity value, can include but is not limited to: rule of thumb configuring The corresponding relationship of key word of the inquiry and complexity value.Alternatively, passing through pair of neural metwork training key word of the inquiry and complexity value It should be related to, with no restrictions to this training process.Alternatively, the inquiry for obtaining the inquiry request is crucial when executing certain inquiry request Word, and the complexity value of the inquiry request is obtained, when such as executing the inquiry request, consume cpu resource, the consumption 100M of 1 core Memory source, then complexity value is the corresponding complexity value of memory source of the cpu resource of 1 core, 100M, is not limited this System.

In one example, inquiry request can include but is not limited to SQL query request；And key word of the inquiry may include But it is not limited to following one or any combination: addition keyword (i.e. join, if SQL query request includes keyword join), Result set is grouped keyword (i.e. groupby, as SQL query request include keyword groupby), to result set into The keyword (i.e. orderby, if SQL query request includes keyword orderby) of row sequence lists different keywords (i.e. Distinct, as SQL query request calculates keyword (i.e. count, as SQL query is asked including keyword distinct), line number It asks including keyword count), window function keyword (i.e. window, if SQL query request includes keyword window).

Ginseng is shown in Table 1, and is an example of the first mapping table, is used for pair of record queries keyword and complexity value It should be related to, complexity value here embodies the complexity of inquiry request.For example, complexity value 5 indicates the CPU of 1 core of consumption Resource, the memory source for consuming 100M, complexity value 10 indicate the cpu resource of 2 core of consumption, consume the memory source of 200M, And so on.Certainly, table 1 is an example, and complexity value corresponding for key word of the inquiry is related with actual conditions, This is repeated no more.

Table 1

Key word of the inquiry	Complexity value
		join	Complexity value 5
groupby	Complexity value 10
		orderby	Complexity value 8
distinct	Complexity value 12
		count	Complexity value 6
window	Complexity value 15

Further, for each inquiry request received in preset time window, in order to obtain looking into for the inquiry request Complexity is ask, can be in the following way: mode one obtains key word of the inquiry from the inquiry request, and closes by the inquiry Key word inquires the first mapping table, to obtain complexity value corresponding with the key word of the inquiry, and the complexity value is determined as this The corresponding inquiry complexity of inquiry request.Mode two obtains key word of the inquiry from multiple subqueries of the inquiry request, and leads to It crosses each key word of the inquiry obtained and inquires the first mapping table, to obtain complexity value corresponding with each key word of the inquiry；So Afterwards, it is complicated that the sum of obtained complexity value (sums of i.e. all complexity values) can be determined as to the corresponding inquiry of the inquiry request Degree.

For example, being directed to mode one, it is assumed that inquiry request is the join sentence of SQL, i.e., the inquiry request includes that inquiry is crucial Word " join " then can obtain complexity value 5, so by the first mapping table shown in key word of the inquiry " join " inquiry table 1 Afterwards, it can determine that the corresponding inquiry complexity of the inquiry request is complexity value 5.

For mode two, it is assumed that inquiry request includes subquery 1, subquery 2, subquery 3, which is SQL Join sentence, the subquery 2 are the groupby sentences of SQL, which is the distinct sentence of SQL.Subquery 1 includes Key word of the inquiry " join " obtains complexity value 5 by the first mapping table shown in key word of the inquiry " join " inquiry table 1；Son Inquiry 2 includes key word of the inquiry " groupby ", by the first mapping table shown in key word of the inquiry " groupby " inquiry table 1, Obtain complexity value 10；Subquery 2 includes key word of the inquiry " distinct ", passes through key word of the inquiry " distinct " inquiry table First mapping table shown in 1 obtains complexity value 12.It is then possible to determine that the corresponding inquiry complexity of the inquiry request is multiple It is between miscellaneous angle value 5, complexity value 10 and complexity value 12 and, i.e., inquiry complexity be complexity value 27.

Three, data volume (being referred to as inquiry scan data volume Query_DataScanned) is inquired, that is, executes inquiry and asks The data volume returned when asking.For example, it is assumed that inquiry request 1 is used for request data A, and the size of data A is 10M, then inquires number It can be 10M according to amount, i.e., be 10M to the data that client returns.

In one example, historical data can be collected, and Data Identification and inquiry data volume are obtained according to historical data Corresponding relationship；Then, the corresponding relationship of the Data Identification Yu the inquiry data volume is recorded in the second mapping table.For example, When executing certain inquiry request, if the inquiry request is used for request data A, and the size of data A is 10M, then front end node can be with It is collected into above- mentioned information (i.e. historical data), and obtains data A and inquires the corresponding relationship of data volume 100, and in the second mapping The corresponding relationship is recorded in table.Ginseng is shown in Table 2, and is an example of the second mapping table, not to the content of this second mapping table It is limited.

Table 2

Data Identification	Inquire data volume
		Data A	10M
Data B	20M

Further, for each inquiry request received in preset time window, in order to obtain looking into for the inquiry request Data volume is ask, it can be in the following way: the second mapping table being inquired by the Data Identification of the inquiry request, is obtained and the data Identify corresponding inquiry data volume.For example, if the Data Identification that the inquiry request carries is data A, it is determined that corresponding with data A Inquiry data volume 10M.If the Data Identification that the inquiry request carries is data C, since the second mapping table does not record C pairs of data The inquiry data volume 10M answered, then the corresponding inquiry data volume of data C can use as default (it can rule of thumb configure, Such as 5M).

Four, query time (being referred to as inquiry short time consumption Query_ResponseTime), i.e. execution inquiry request When consumed time (from inquiry request is started to process to inquiry request time consumed by processing terminate).For example, it is assumed that holding When row inquiry request 1,3 seconds are consumed altogether, then query time is 3 seconds.

Wherein it is possible to collect historical data, the corresponding relationship of Data Identification and query time is obtained according to historical data, The corresponding relationship of the Data Identification and query time is recorded in second mapping table.It is looked into for each of being received in preset time window Request is ask, in order to obtain the query time of the inquiry request, in the following way: passing through the Data Identification of inquiry request inquiry the Two mapping tables obtain query time corresponding with the Data Identification.

Five, when resources occupation rate (also referred to as resource utilization Resource_Utilization), i.e. execution inquiry request Consumed resource, such as memory usage, CPU usage, network bandwidth occupancy.Assuming that when executing inquiry request 1, consumption The cpu resource of 1 core, the memory source of 100M, 100M network bandwidth, then resources occupation rate be 1 core cpu resource, The network bandwidth of the memory source of 100M, 100M.

Wherein it is possible to obtain Data Identification and resources occupation rate by collecting historical data, and according to the historical data Corresponding relationship.It is then also possible to record the corresponding relationship of the Data Identification Yu the resources occupation rate in the second mapping table.Into One step, for each inquiry request received in preset time window, in order to obtain the resources occupation rate of the inquiry request, also It can be in the following way: second mapping table being inquired by the Data Identification of the inquiry request, so as to obtain and be somebody's turn to do The corresponding resources occupation rate of Data Identification.

In one example, front end node can be with the second mapping table shown in Maintenance Table 3, and second mapping table is for remembering The corresponding relationship recorded Data Identification, inquire data volume, query time, resources occupation rate.It is inscribed for preset time window based on this The each inquiry request received, can the second mapping table shown in the Data Identification inquiry table 3 by the inquiry request, thus To characteristic information corresponding with the Data Identification, this feature information may include inquiry data volume, query time, resources occupation rate In one or more.

Table 3

Data Identification	Inquire data volume	Query time	Resources occupation rate
				Data A	10M	3 seconds	Cpu resource: 1 core；Memory source: 100M；Network bandwidth: 100M
Data B	20M	6 seconds	Cpu resource: 2 cores；Memory source: 200M；Network bandwidth: 200M

In conclusion if the Data Identification that above-mentioned inquiry request carries is data A, it is determined that inquiry corresponding with data A Data volume 10M, query time 3 seconds, resources occupation rate " cpu resource: 1 core；Memory source: 100M；Network bandwidth: 100M ".This It outside,, then can be with since the second mapping table does not record the corresponding content of data C if the Data Identification that inquiry request carries is data C Inquiry data volume is used as default and query time uses as default and resources occupation rate uses as default, It is without limitation.

By the above process, the characteristic information of each inquiry request received in available preset time window, with spy Reference breath is concurrency, inquiry complexity, inquires for data volume, query time, resources occupation rate.

In step 201, resource overhead is obtained according to the characteristic information of the inquiry request received, may include: to be directed to The each inquiry request received in preset time window, the prediction of the inquiry request is obtained according to the characteristic information of the inquiry request Stock number, and resource overhead is determined according to the prognostic resources of each inquiry request, for example, resource overhead can be each inquiry The sum of prognostic resources of request.

Wherein, when the characteristic information according to the inquiry request obtains the prognostic resources of the inquiry request, it is assumed that the spy Reference breath for inquiry complexity, then inquire complexity complexity value it is bigger when, prognostic resources are bigger, inquire answering for complexity Miscellaneous angle value is got over hour, and prognostic resources are smaller, with no restrictions to this determination process, as long as meeting above-mentioned rule.Assuming that should Characteristic information be inquiry data volume, then inquire data volume it is bigger when, prognostic resources are bigger, inquiry data volume get over hour, predict Stock number is smaller, with no restrictions to this determination process, as long as meeting above-mentioned rule.Assuming that when this feature information is inquiry Between, then when query time is bigger, prognostic resources are bigger, and query time is got over hour, and prognostic resources are smaller, to this determination process With no restrictions, as long as meeting above-mentioned rule.Assuming that this feature information is resources occupation rate, then when resources occupation rate is bigger, Prognostic resources are bigger, and resources occupation rate is got over hour, and prognostic resources are smaller, with no restrictions to this determination process, as long as meeting Above-mentioned rule.Certainly, at least one example of aforesaid way, it is without limitation.

For example, when this feature information is concurrency, in inquiry complexity, inquiry data volume, query time, resources occupation rate It is multiple when, for including this 5 features, then can to concurrency, inquiry complexity, inquiry data volume, query time, money Source occupancy is normalized, i.e., normalizes concurrency, inquiry complexity, inquiry data volume, query time, resources occupation rate To same number of levels, with no restrictions to this normalization mode.Assuming that concurrency A after being normalized, inquiry complexity B, Inquire data quantity C, query time D, resources occupation rate E, then can to concurrency A, inquiry complexity B, inquiry data quantity C, look into Inquiry time D, resources occupation rate E sum, if summed result is bigger, prognostic resources are bigger, and summed result is got over hour, in advance It is smaller to survey stock number, it is without limitation, as long as meeting above-mentioned rule.

In another example can also be to (weight 1* concurrency A), (weight 2* inquires complexity B), (weight 3* inquiry data volume C), (weight 4* query time D), (weight 5* resources occupation rate E) sum, if summed result is bigger, prognostic resources Bigger, summed result is got over hour, and prognostic resources are smaller, without limitation, as long as meeting above-mentioned rule.Wherein, it weighs Weighing 1, weight 2, weight 3, weight 4 and weight 5 can rule of thumb configure, without limitation.For example, weight 1, weight 2, Weight 3, weight 4, weight 5 and can be 1, naturally it is also possible to be other numerical value, such as 2,3.

In one example, the prognostic resources of the inquiry request are obtained according to the characteristic information of the inquiry request, it can be with Include: to be analyzed by characteristic information of the prediction model to the inquiry request, obtains the prognostic resources of the inquiry request；It should Prediction model can include but is not limited to: Holt-Winter (third index flatness) seaconal model, ARMA (autoregression and cunning It is dynamic average) model, linear regression model (LRM), neural network model.

By taking prediction model is neural network model as an example, then neural network can use historical data training characteristics information with The corresponding relationship of prognostic resources.For example, can then train inquiry complexity and prediction to provide when characteristic information is inquiry complexity The corresponding relationship of source amount.For example, when executing some inquiry request, it is assumed that it is complexity value 5, actual consumption that it, which inquires complexity, Stock number be stock number A, then the corresponding relationship of available complexity value 5 and prognostic resources A, certainly, neural network is By the corresponding relationship of a large amount of historical datas training inquiry complexity and prognostic resources, with no restrictions to this training process, In training result, the complexity value for inquiring complexity is bigger, and prognostic resources are bigger, and the complexity value for inquiring complexity is smaller, Prognostic resources are smaller.For the characteristic informations such as concurrency, inquiry data volume, query time, resources occupation rate, training process Similar, details are not described herein.When this feature information is concurrency, inquiry complexity, inquiry data volume, query time, resource account for When with multiple in rate, training process is similar, and details are not described herein.

Further, after the corresponding relationship that neural metwork training goes out characteristic information and prognostic resources, for it is default when Between each inquiry request for receiving in window, neural network can inquire described corresponding close according to the characteristic information of the inquiry request System, obtains the prognostic resources of the inquiry request, with no restrictions to this process.

Certainly, aforesaid way obtains an example of prognostic resources just with neural network model, does not limit this System.When prediction model is Holt-Winter seaconal model, arma modeling, linear regression model (LRM), implementation and nerve net The implementation of network model is similar, and it is no longer repeated herein.In short, as long as determination process meets following rule: inquiry When the complexity value of complexity is bigger, prognostic resources are bigger；When inquiry data volume is bigger, prognostic resources are bigger；When inquiry Between it is bigger when, prognostic resources are bigger；When resources occupation rate is bigger, prognostic resources are bigger；When number of concurrent is bigger, resource is predicted It measures bigger.

It in step 202, can be with according to the calculate node in resource overhead and calculate node resource dynamic adjustment resource pond Including but not limited to: calculate node quantity is obtained according to resource overhead and calculate node resource；It is then possible to divide in resource pool With the calculate node with the calculate node quantity Matching.

Wherein, calculate node quantity is obtained according to the resource overhead and calculate node resource, can include but is not limited to as Under type: it rounds up to the resource overhead/calculate node resource, it can obtain calculate node quantity.It is, of course, also possible to adopt Calculate node quantity is otherwise obtained, as long as calculate node quantity is upward more than or equal to resource overhead/calculate node resource The result of rounding, it is without limitation.

For example, when the sum of prognostic resources of all inquiry requests received in preset time window are 100 CPU cores, When i.e. resource overhead is 100 CPU cores, it is assumed that calculate node resource is 8 CPU core (each calculate nodes i.e. in resource pool It is provided which the calculate node resource of 8 CPU cores), then calculate node quantity can be 13.Obviously, when calculate node quantity is At 13, since 13 calculate nodes can provide 104 CPU cores, 13 calculate nodes can satisfy 100 CPU cores Resource overhead, that is to say, that 13 calculate nodes are capable of handling all inquiry requests received in preset time window.

In another example when resource overhead is 20G memory, it is assumed that calculate node resource is 2G memory, then calculate node quantity It can be 10.Obviously, when calculate node quantity is 10, since 10 calculate nodes can provide 20G memory, 10 calculate nodes can satisfy the resource overhead of 20G memory, that is to say, that 10 calculate nodes are capable of handling preset time window All inquiry requests inside received.

In another example when resource overhead is 100 CPU cores, 20G memory, calculate node resource is 8 CPU cores, 2G memory When, then CPU core resource is needed using 13 calculate nodes, and memory source is needed using 10 calculate nodes, therefore, can will most Big calculate node quantity 13, is determined as calculate node quantity.

Wherein, the calculate node with the calculate node quantity Matching is distributed in resource pool, if may include: in resource pool The quantity of already existing calculate node be less than the calculate node quantity, then can in resource pool dilatation calculate node so that The quantity of calculate node after dilatation is more than or equal to the calculate node quantity.If the number of already existing calculate node in resource pool Amount be greater than the calculate node quantity, then can in resource pool capacity reducing calculate node so that the quantity of the calculate node after capacity reducing More than or equal to the calculate node quantity.

For example, it is assumed that have existed 8 calculate nodes in resource pool, and above-mentioned calculate node quantity is 13, then can be with New 5 calculate nodes of dilatation in resource pool, in this way, one 13 calculate nodes are co-existed in resource pool, and this 13 calculating sections Point is for handling all inquiry requests received in preset time window.

In another example, it is assumed that 20 calculate nodes are had existed in resource pool, and above-mentioned calculate node quantity is 13, then may be used With 7 calculate node of capacity reducing in resource pool, in this way, one 13 calculate nodes are co-existed in resource pool, and this 13 calculate nodes For handling all inquiry requests received in preset time window.

In one example, front end node can be sent after obtaining calculate node quantity 13 to resource allocation server Carry the scalable appearance order of resource of calculate node quantity 13.Resource allocation server after receiving the resource scalable appearance order, It can be distributed in resource pool and the matched calculate node of calculate node quantity 13.

For example, a front end node if it exists, then resource allocation server, which receives only, carries calculate node quantity 13 The scalable appearance order of resource, therefore, dilatation/capacity reducing calculate node in resource pool, so that there are 13 calculate nodes in resource pool. In another example two front end nodes if it exists, it is assumed that resource allocation server, which receives, to be carried the resource of calculate node quantity 13 and expand Capacity reducing order, the scalable appearance of the resource for carrying calculate node quantity 8 are ordered, then dilatation/capacity reducing calculate node in resource pool, so that There are 21 calculate nodes in resource pool.

Wherein, for resource allocation server in resource pool when dilatation/capacity reducing calculate node, performance can be second grade (even Hundred Milliseconds can be optimized to), i.e., only need several seconds clock time (or even hundred Milliseconds can be optimized to), so that it may expand in resource pool Hold calculate node or capacity reducing calculate node.

In step 203, data corresponding with above-mentioned inquiry request are inquired by the calculate node in resource pool, can wrap Include: for each inquiry request received in preset time window, front end node can carry out SQL parsing to the inquiry request, Inquiry request is generated using SQL parsing result, and the inquiry request is sent to calculate node；Calculate node is receiving inquiry After request, data corresponding with the inquiry request can be read from data source and are calculated, and return data to preceding end segment Point；The data received are returned to client by front end node.For example, inquiry request is split into 6 subqueries by front end node Request, with no restrictions to this process, and by 6 sub- inquiry request load balancing to 6 calculate nodes.For each calculate node For, after calculate node receives subquery request, corresponding with subquery request data are read from data source, and by data Return to front end node.Front end node combines these data one after receiving the data for 6 sub- inquiry requests It rises, obtains data acquisition system, and the data acquisition system after combining is exactly the corresponding data of above-mentioned inquiry request.Then, by the data set Conjunction is sent to client, is finally completed data query operation.

Based on the above-mentioned technical proposal, it in the embodiment of the present application, can be obtained according to the characteristic information of the inquiry request received Resource overhead is obtained, and calculate node quantity is obtained according to resource overhead and calculate node resource, and distributes and is somebody's turn to do in resource pool The calculate node of calculate node quantity Matching.In this way, can be with the calculate node in dynamic adjustment resource pond, so that in resource pool Calculate node is capable of handling all inquiry requests received, more effectively improves treatment effeciency and the utilization of resources of calculate node Rate can enable calculate node more effectively parallel processing multiple queries to request, improve cpu resource, memory source, network The utilization rate of bandwidth resources improves to reach a better effect from overall calculation resource and user query angles of loading User's use feeling.It is analyzed and predicted by the feature to inquiry request, intelligence can be carried out to the resource of calculate node Analysis and adjust automatically more effectively improve the resource utilization and cost performance of cloud database and cloud data analysis service cluster. Moreover, allowing each calculate node to provide serverless backup for user by the calculate node in dynamic adjustment resource pond (Serverless) query analysis service need to only perceive cloud service so that user is not necessarily to aware services device or Service Instance Service of offer itself, is based on cloud service, and user only needs to input SQL query request, so that it may by calculate node in database Middle progress data query and analysis, can be with Seamless integration- business analysis tool and application program.

It is shown in Figure 3, it is the another application schematic diagram of a scenario of the embodiment of the present application, below to the difference of Fig. 3 and Fig. 1 Place is illustrated.In Fig. 1, all calculate nodes are all located at the same resource pool, can be by the money of calculate node in Fig. 3 Source pond is divided into multiple child resource ponds, and by taking child resource pond 1, child resource pond 2, child resource pond 3 as an example, and calculate node is to be located at son Resource pool.For example, child resource pond 1 includes 2 calculate nodes, child resource pond 2 includes 2 calculate nodes, and child resource pond 3 includes 4 A calculate node is that dilatation or capacity reducing processing are carried out to the calculate node of sub- resource pool, rather than for money in the present embodiment Source pond.

For the same child resource pond, the calculate node resource of all calculate nodes is identical；For different child resource ponds, The calculate node resource of calculate node can be same or different.For example, the calculate node money of the calculate node in child resource pond 1 Source is 4 CPU cores, and the calculate node resource of the calculate node in child resource pond 2 is 8 CPU cores, the calculating in child resource pond 3 The calculate node resource of node is 16 CPU cores.

Wherein it is possible to the child resource pond of different stage be divided for different user, for example, can according to the demand of different user With SLA (the Service-Level Agreement, between service-level agreement, i.e. Internet service provider and user based on user A contract, define the terms such as service type, service quality and customer payment) information, for different user divide it is not at the same level Other child resource pond, to meet the needs of different user.

It is shown in Figure 4 under above-mentioned application scenarios, for the process of the data query method proposed in the embodiment of the present application Schematic diagram is applied to for front end node in this way, and this method may comprise steps of:

Step 401, according to the characteristic information of the inquiry request received, the inquiry request received is divided at least one A distribution group；Different distribution groups corresponds to different child resource ponds.As according to the inquiry request received in preset time window The inquiry request received is divided at least one distribution group by characteristic information.

Step 402, the resource overhead of the distribution group is obtained according to the characteristic information of the inquiry request in distribution group.

Step 403, according to the calculate node resource of the resource overhead of the distribution group and the corresponding child resource pond of the distribution group, Calculate node in the dynamic regulation child resource pond.

Step 404, number corresponding with the inquiry request in the distribution group is inquired by the calculate node in the child resource pond According to, that is to say, that different inquiry requests may be assigned to the calculate node in different child resource ponds.

Before executing step 401, for all inquiry requests received, each inquiry request can also be first obtained Characteristic information, this feature information can include but is not limited to following one or any combination: concurrency, is looked into inquiry complexity Ask data volume, query time, resources occupation rate.Wherein, for the acquisition modes of characteristic information, it may refer to stream shown in Fig. 2 Journey, it is no longer repeated herein.

In step 401, according to the characteristic information of the inquiry request received, by the inquiry request received be divided into A few distribution group, can include but is not limited to:, can be according to the spy of the inquiry request for each inquiry request received The prognostic resources of the information acquisition inquiry request are levied, and determine resource section belonging to the prognostic resources, and by the inquiry Request is divided into the corresponding distribution group in the resource section；Wherein, different distribution groups can correspond to different resource sections.

Wherein, the process for obtaining the prognostic resources of inquiry request may refer to step 201, and details are not described herein.

Wherein it is determined that resource section belonging to the prognostic resources, and the inquiry request is divided into the resource section pair The distribution group answered, can include but is not limited to: configuring resource section in advance for each child resource pond, does not limit this configuration mode System, for example, the resource section in the child resource pond can be bigger when the calculate node resource of group resource pool is bigger, works as child resource The calculate node resource in pond gets over hour, and the resource section in the child resource pond can be smaller.For example, the calculate node in child resource pond 1 Resource is 4 CPU cores, and the calculate node resource in child resource pond 2 is 8 CPU cores, and the calculate node resource in child resource pond 3 is 16 A CPU core, then the resource section in child resource pond 1 be [0-1) a CPU core, the resource section in child resource pond 2 be [1-2) a CPU Core, the resource section in child resource pond 3 be [2- is infinitely great) a CPU core.Further, it is also possible to configure one point for each resource section Combo, for example the resource section in child resource pond 1 configure distribution group 1, are that the resource section in child resource pond 2 configures distribution group 2, for son The resource section of resource pool 3 configures distribution group 3.Obviously, the corresponding child resource pond 1 of distribution group 1, the corresponding child resource pond 2 of distribution group 2, Distribution group 3 corresponds to child resource pond 3.

Further, it is assumed that the prognostic resources of inquiry request are 1 CPU core, then can determine the prognostic resources institute The resource section of category is the resource section in child resource pond 2, and the inquiry request can be divided into distribution group 2.Obviously, to pre- If these inquiry requests can be divided into each point after all inquiry requests received in time window carry out above-mentioned processing Combo, if inquiry request 1-10 is divided into distribution group 1, inquiry request 11-50 is divided into distribution group 2, inquiry request 51- 100 are divided into distribution group 3.

In step 402, the resource overhead that the distribution group is obtained according to the characteristic information of the inquiry request in distribution group, can To include: to obtain the pre- of the inquiry request according to the characteristic information of the inquiry request for each inquiry request in distribution group Stock number is surveyed, and obtains the resource overhead of distribution group according to the prognostic resources.

Wherein, the realization process of step 402 may refer to step 201, the difference is that: it is to be directed in step 201 All inquiry requests received are handled, and in step 402, it is at all inquiry requests in distribution group Reason, and other processes are similar, it is no longer repeated herein.

In step 403, it is provided according to the calculate node in the resource overhead of distribution group and the corresponding child resource pond of the distribution group Source, the calculate node in child resource pond described in dynamic regulation, may include: the resource overhead and the distribution group pair according to distribution group The calculate node resource in the child resource pond answered obtains the calculate node quantity in the child resource pond；It is distributed in the child resource pond With the calculate node of the calculate node quantity Matching.

Further, the calculate node with the calculate node quantity Matching is distributed in the child resource pond, if may include: The quantity of already existing calculate node is less than the calculate node quantity in the child resource pond, then the dilatation meter in the child resource pond The quantity of operator node, the calculate node after dilatation is more than or equal to calculate node quantity；If already existing meter in the child resource pond The quantity of operator node is greater than calculate node quantity, then the capacity reducing calculate node in the child resource pond, the calculate node after capacity reducing Quantity is more than or equal to the calculate node quantity.

Wherein, the realization process of step 403 may refer to step 202, the difference is that: it is basis in step 202 The resource overhead and calculate node resource of all inquiry requests received, the calculate node in dynamic adjustment resource pond, and walk It is the calculate node resource of the resource overhead and the corresponding child resource pond of the distribution group according to distribution group, dynamic regulation in rapid 403 Calculate node in the child resource pond.

For example, be directed to step 403, can according to the resource overhead of distribution group 1 and the calculate node resource in child resource pond 1, The calculate node quantity 10 in child resource pond 1 is obtained, and distributes 10 calculate nodes in child resource pond 1.Furthermore, it is possible to according to The resource overhead of distribution group 2 and the calculate node resource in child resource pond 2 obtain the calculate node quantity 8 in child resource pond 2, and 8 calculate nodes are distributed in child resource pond 1.Furthermore, it is possible to according to the calculating of the resource overhead of distribution group 3 and child resource pond 3 Node resource obtains the calculate node quantity 13 in child resource pond 3, and 13 calculate nodes are distributed in child resource pond 3.

Wherein, the realization process of step 404 may refer to step 203, the difference is that: it is that will look into step 203 It askes the calculate node for requesting corresponding inquiry request to be sent to resource pool, is by the inquiry request pair of distribution group 1 in step 404 The inquiry request answered is sent to the calculate node in child resource pond 1, and the corresponding inquiry request of the inquiry request of distribution group 2 is sent to The corresponding inquiry request of the inquiry request of distribution group 3 is sent to the calculating section in child resource pond 3 by the calculate node in child resource pond 2 Point, it is no longer repeated herein.

Based on similarly applying conceiving with the above method, the embodiment of the present application also provides a kind of data query device, such as Fig. 5 Shown, for the structure chart of the device, which includes:

Module 501 is obtained, for obtaining resource overhead according to the characteristic information of the inquiry request received；Processing module 502, according to the calculate node in resource overhead and calculate node resource dynamic adjustment resource pond；Enquiry module 503, for passing through The calculate node inquires data corresponding with the inquiry request.

In one example, the acquisition module 501 is also used to: when characteristic information includes inquiry complexity, from inquiry Key word of the inquiry is obtained in request；The first mapping table is inquired by the key word of the inquiry, is obtained and the key word of the inquiry pair The complexity value is determined as the corresponding inquiry complexity of the inquiry request by the complexity value answered；Alternatively, from inquiry request Multiple subqueries in obtain key word of the inquiry；The first mapping table is inquired by the key word of the inquiry of acquisition, obtains closing with inquiry The corresponding complexity value of key word；The sum of obtained complexity value is determined as the corresponding inquiry complexity of the inquiry request；Its In, first mapping table is used for the corresponding relationship of record queries keyword and complexity value.

Based on similarly conceiving with the above method, the embodiment of the present application provides a kind of data query equipment, including processor, For obtaining resource overhead according to the characteristic information of the inquiry request received；According to the resource overhead and calculate node resource Calculate node in dynamic adjustment resource pond；Data corresponding with the inquiry request are inquired by the calculate node.

Based on similarly applying conceiving with the above method, the embodiment of the present application also provides a kind of machine readable storage medium, It can be applied to data query equipment, several computer instructions be stored on machine readable storage medium；Wherein, the computer Instruction, which is performed, to be handled as follows: obtaining resource overhead according to the characteristic information of the inquiry request received；According to described Calculate node in resource overhead and calculate node resource dynamic adjustment resource pond；It is inquired by the calculate node and is looked into described It askes and requests corresponding data.

Based on similarly applying conceiving with the above method, the embodiment of the present application also provides a kind of data query device, such as Fig. 6 Shown, for the structure chart of the device, which includes:

Division module 601 divides the inquiry request received for the characteristic information according to the inquiry request received To at least one distribution group；Different distribution groups correspond to different child resource ponds；Module 602 is obtained, for according to looking into distribution group The characteristic information for asking request obtains the resource overhead of the distribution group；Processing module 603, for the resource according to the distribution group The calculate node resource of expense and the corresponding child resource pond of the distribution group, the calculating section in child resource pond described in dynamic regulation Point；Enquiry module 604, for passing through the calculate node inquiry in the child resource pond and the inquiry request pair in the distribution group The data answered.

In one example, the division module 603 is specifically used for: for the inquiry request received, according to the inquiry The characteristic information of request obtains the prognostic resources of the inquiry request, and determines resource section belonging to the prognostic resources；It will The inquiry request is divided into the corresponding distribution group in the resource section；Wherein, different distribution groups corresponds to different resource sections.

Based on similarly conceiving with the above method, the embodiment of the present application provides a kind of data query equipment, including processor, For the characteristic information according to the inquiry request received, the inquiry request received is divided at least one distribution group；Its In, different distribution groups corresponds to different child resource ponds；Described point is obtained according to the characteristic information of the inquiry request in distribution group The resource overhead of combo；It is provided according to the calculate node in the resource overhead of the distribution group and the corresponding child resource pond of the distribution group Source, the calculate node in child resource pond described in dynamic regulation；Pass through the calculate node inquiry and the distribution group in child resource pond In the corresponding data of inquiry request.

Based on similarly applying conceiving with the above method, the embodiment of the present application also provides a kind of machine readable storage medium, It can be applied to data query equipment, several computer instructions be stored on machine readable storage medium；Wherein, the computer Instruction, which is performed, to be handled as follows: according to the characteristic information of the inquiry request received, the inquiry request received being drawn Assign at least one distribution group；Wherein, different distribution groups corresponds to different child resource ponds；According to the inquiry request in distribution group Characteristic information obtain the resource overhead of the distribution group；It is corresponding according to the resource overhead of the distribution group and the distribution group The calculate node resource in child resource pond, the calculate node in child resource pond described in dynamic regulation；Pass through the calculating in child resource pond Querying node data corresponding with the inquiry request in the distribution group.

System, device, module or the unit that above-described embodiment illustrates can specifically realize by computer chip or entity, Or it is realized by the product with certain function.A kind of typically to realize that equipment is computer, the concrete form of computer can To be personal computer, laptop computer, cellular phone, camera phone, smart phone, personal digital assistant, media play In device, navigation equipment, E-mail receiver/send equipment, game console, tablet computer, wearable device or these equipment The combination of any several equipment.

For convenience of description, it is divided into various units when description apparatus above with function to describe respectively.Certainly, implementing this The function of each unit can be realized in the same or multiple software and or hardware when application.

It should be understood by those skilled in the art that, embodiments herein can provide as method, system or computer program Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the application Apply the form of example.Moreover, it wherein includes computer usable program code that the embodiment of the present application, which can be used in one or more, The computer implemented in computer-usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) The form of program product.

The application is referring to method, the process of equipment (system) and computer program product according to the embodiment of the present application Figure and/or block diagram describe.It is generally understood that being realized by computer program instructions each in flowchart and/or the block diagram The combination of process and/or box in process and/or box and flowchart and/or the block diagram.It can provide these computer journeys Sequence instruct to general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices processor with A machine is generated, so that the instruction generation executed by computer or the processor of other programmable data processing devices is used for Realize the dress for the function of specifying in one or more flows of the flowchart and/or one or more blocks of the block diagram It sets.

Moreover, these computer program instructions also can store be able to guide computer or other programmable datas processing set In standby computer-readable memory operate in a specific manner, so that instruction stored in the computer readable memory generates Manufacture including command device, the command device are realized in one process of flow chart or multiple processes and/or block diagram one The function of being specified in a box or multiple boxes.

These computer program instructions can also be loaded into computer or other programmable data processing devices, so that counting Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer Or the instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram The step of function of being specified in one box or multiple boxes.

The above description is only an example of the present application, is not intended to limit this application.For those skilled in the art For, various changes and changes are possible in this application.All any modifications made within the spirit and principles of the present application are equal Replacement, improvement etc., should be included within the scope of the claims of this application.

Claims

1. a kind of data query method, which is characterized in that the described method includes:

Data corresponding with the inquiry request are inquired by the calculate node.

2. the method according to claim 1, wherein the characteristic information includes following one or any group It closes: concurrency, inquiry complexity, inquiry data volume, query time, resources occupation rate.

3. the method according to claim 1, wherein the characteristic information for the inquiry request that the basis receives obtains Before obtaining resource overhead, the method also includes:

If the characteristic information includes inquiry complexity, key word of the inquiry is obtained from inquiry request；

The first mapping table is inquired by the key word of the inquiry, obtains complexity value corresponding with the key word of the inquiry, and will The complexity value is determined as the corresponding inquiry complexity of the inquiry request；

Wherein, first mapping table is used for the corresponding relationship of record queries keyword and complexity value.

4. the method according to claim 1, wherein the characteristic information for the inquiry request that the basis receives obtains Before obtaining resource overhead, the method also includes:

If the characteristic information includes inquiry complexity, key word of the inquiry is obtained from multiple subqueries of inquiry request；It is logical It crosses the key word of the inquiry obtained and inquires the first mapping table, obtain complexity value corresponding with key word of the inquiry；The complexity that will be obtained The sum of angle value is determined as the corresponding inquiry complexity of the inquiry request；

5. the method according to claim 3 or 4, which is characterized in that

The inquiry request includes: the request of structured query language SQL query；The key word of the inquiry include following one or Any combination: keyword is added, the keyword that result set is grouped, the keyword that result set is ranked up, lists not Same keyword, line number calculate keyword, window function keyword.

6. the method according to claim 1, wherein the characteristic information for the inquiry request that the basis receives obtains Before obtaining resource overhead, the method also includes:

The second mapping table is inquired by the Data Identification of inquiry request, obtains characteristic information corresponding with the Data Identification；Its In, second mapping table is used to record the corresponding relationship of Data Identification and characteristic information；The characteristic information includes inquiry number According to one or more in amount, query time, resources occupation rate.

7. according to the method described in claim 6, it is characterized in that, described reflected by the Data Identification inquiry second of inquiry request Firing table, before obtaining characteristic information corresponding with the Data Identification, further includes:

Historical data is collected, the corresponding relationship of Data Identification and characteristic information is obtained according to the historical data；

The corresponding relationship of Data Identification and characteristic information is recorded in second mapping table.

8. the method according to claim 1, wherein

The characteristic information for the inquiry request that the basis receives obtains resource overhead, comprising:

For the inquiry request received, the prognostic resources of the inquiry request are obtained according to the characteristic information of the inquiry request, And resource overhead is determined according to the prognostic resources of inquiry request.

9. according to the method described in claim 8, it is characterized in that,

The characteristic information according to the inquiry request obtains the prognostic resources of the inquiry request, comprising:

It is analyzed by characteristic information of the prediction model to the inquiry request, obtains the prognostic resources of the inquiry request；Its In, the prediction model includes: third index flatness Holt-Winter seaconal model, autoregression and sliding average ARMA mould Type, linear regression model (LRM), neural network model.

10. the method according to claim 1, wherein described according to the resource overhead and calculate node resource Calculate node in dynamic adjustment resource pond, comprising:

Calculate node quantity is obtained according to the resource overhead and calculate node resource；

The calculate node with the calculate node quantity Matching is distributed in resource pool.

11. according to the method described in claim 10, it is characterized in that,

The calculate node distributed in resource pool with the calculate node quantity Matching, comprising:

If the quantity of already existing calculate node is less than the calculate node quantity, the dilatation meter in resource pool in resource pool The quantity of operator node, the calculate node after dilatation is more than or equal to the calculate node quantity；

If the quantity of already existing calculate node is greater than the calculate node quantity, the capacity reducing meter in resource pool in resource pool The quantity of operator node, the calculate node after capacity reducing is more than or equal to the calculate node quantity.

12. a kind of data query method, which is characterized in that the described method includes:

According to the characteristic information of the inquiry request received, the inquiry request received is divided at least one distribution group；Its In, different distribution groups corresponds to different child resource ponds；

According to the calculate node resource of the resource overhead of the distribution group and the corresponding child resource pond of the distribution group, dynamic regulation Calculate node in the child resource pond；

13. according to the method for claim 12, which is characterized in that the characteristic information includes following one or any group It closes: concurrency, inquiry complexity, inquiry data volume, query time, resources occupation rate.

14. according to the method for claim 12, which is characterized in that

The inquiry request that will be received is divided into before at least one distribution group, the method also includes:

15. according to the method for claim 12, which is characterized in that

16. method according to claim 14 or 15, which is characterized in that

17. according to the method for claim 12, which is characterized in that

18. according to the method for claim 17, which is characterized in that the Data Identification by inquiry request inquires second Mapping table, before obtaining characteristic information corresponding with the Data Identification, further includes:

19. according to the method for claim 12, which is characterized in that, will according to the characteristic information of the inquiry request received The inquiry request received is divided at least one distribution group, comprising:

For the inquiry request received, the prognostic resources of the inquiry request are obtained according to the characteristic information of the inquiry request, And determine resource section belonging to the prognostic resources；The inquiry request is divided into the corresponding distribution group in the resource section； Wherein, different distribution groups corresponds to different resource sections.

20. according to the method for claim 12, which is characterized in that

The resource overhead of the distribution group is obtained according to the characteristic information of the inquiry request in distribution group, comprising:

For the inquiry request in distribution group, the prediction resource of the inquiry request is obtained according to the characteristic information of the inquiry request It measures, and obtains the resource overhead of the distribution group according to the prognostic resources.

21. method described in 9 or 20 according to claim 1, which is characterized in that

22. according to the method for claim 12, which is characterized in that

It is described according to the resource overhead of the distribution group and the calculate node resource in the corresponding child resource pond of the distribution group, dynamic Adjust the calculate node in the child resource pond, comprising:

According to the calculate node resource of the resource overhead of the distribution group and the corresponding child resource pond of the distribution group, described in acquisition Calculate node quantity in child resource pond；

The calculate node with the calculate node quantity Matching is distributed in the child resource pond.

23. according to the method for claim 22, which is characterized in that

The calculate node with the calculate node quantity Matching is distributed in the child resource pond, comprising:

If the quantity of already existing calculate node is less than the calculate node quantity in child resource pond, in child resource Chi Zhongkuo Hold calculate node, the quantity of the calculate node after dilatation is more than or equal to the calculate node quantity；

If the quantity of already existing calculate node is greater than the calculate node quantity in child resource pond, contract in child resource pond Hold calculate node, the quantity of the calculate node after capacity reducing is more than or equal to the calculate node quantity.

24. a kind of data query device, which is characterized in that described device includes:

25. device according to claim 24, which is characterized in that the acquisition module is also used to: when characteristic information includes When inquiring complexity, key word of the inquiry is obtained from inquiry request；The first mapping table is inquired by the key word of the inquiry, is obtained It is complicated to be determined as the corresponding inquiry of the inquiry request by complexity value corresponding with the key word of the inquiry for the complexity value Degree；Alternatively, obtaining key word of the inquiry from multiple subqueries of inquiry request；It is reflected by the key word of the inquiry inquiry first of acquisition Firing table obtains complexity value corresponding with key word of the inquiry；The sum of obtained complexity value is determined as the inquiry request pair The inquiry complexity answered；Wherein, first mapping table is used for the corresponding relationship of record queries keyword and complexity value.

26. a kind of data query device, which is characterized in that be applied to front end node, described device includes:

The inquiry request received is divided at least by division module for the characteristic information according to the inquiry request received One distribution group；Wherein, different distribution groups correspond to different child resource ponds；

Module is obtained, for obtaining the resource overhead of the distribution group according to the characteristic information of the inquiry request in distribution group；

Processing module, for according to the resource overhead of the distribution group and the calculate node in the corresponding child resource pond of the distribution group Resource, the calculate node in child resource pond described in dynamic regulation；

Enquiry module, it is corresponding with the inquiry request in the distribution group for being inquired by the calculate node in the child resource pond Data.

27. device according to claim 26, which is characterized in that the division module is specifically used for: for what is received Inquiry request obtains the prognostic resources of the inquiry request according to the characteristic information of the inquiry request, and determines the prediction resource Resource section belonging to amount；The inquiry request is divided into the corresponding distribution group in the resource section；Wherein, different distribution group Corresponding different resource section.

28. a kind of data query equipment characterized by comprising

Processor, for obtaining resource overhead according to the characteristic information of the inquiry request received；According to the resource overhead and Calculate node in calculate node resource dynamic adjustment resource pond；It is inquired by the calculate node corresponding with the inquiry request Data.

29. a kind of data query equipment characterized by comprising processor, for the feature according to the inquiry request received The inquiry request received is divided at least one distribution group by information；Wherein, different distribution groups corresponds to different child resources Pond；The resource overhead of the distribution group is obtained according to the characteristic information of the inquiry request in distribution group；According to the distribution group The calculate node resource of resource overhead and the corresponding child resource pond of the distribution group, the calculating in child resource pond described in dynamic regulation Node；Data corresponding with the inquiry request in the distribution group are inquired by the calculate node in child resource pond.